Evolution and Protein Folds..Gene duplication

Graeme Wistow graeme at MGE2.NEI.NIH.GOV
Fri Jul 1 10:14:18 EST 1994


On 30 Jun 1994, Dan Weinreich wrote:
in reply to my post pointing out that 
" New function -> adaptive conflict -> gene duplication"
is a possible model for gene duplication as illustrated by the
recruitment of crystallins in the eye lens.

:Let's be clear: I presume you don't mean that an "adaptive
:conflict"
:CAUSES a duplication event; only that the advantage of the
:duplication of
:a gene whose product is under such conflict makes its persistance
:more
:likely.  Nevertheless, you acknowledge that the duplication itself
:can
:only be a chance event. 

Of course.

> .... This seems to have happened in the case of
delta-crystallin2... 

:Could you describe the data supporting this assertion?  How does
:it assign
:temporal order to the two events (add'n of 2nd function to single
:gene
:product, and gene duplication event).

Well its hard to summarize the whole crystallin field here. It has
been extensively reviewed (some might say over-reviewed).
A general overview with some useful references was in TIBS, 18,
301-306 (1993).
However...
First there is ample evidence of the acquisition of a new function
without gene duplication. epsilon-crystallin is identical to
lactate dehydrogenase B; tau-crystallin is identical to alpha-
enolase; zeta-crystallin is identical to a quinone reductase. You
could take a look at Lee et al, (1994), J. Mol. Biol. 236,669-678
and references therein. Gene duplication is not necessary before an
enzyme becomes a crystallin.

The idea that an unduplicated enzyme crystallin is under separate
and perhaps competing selective pressures come from observations
like the replacement of Phe 118 by glycine in LDHB only when that
enzyme is also serving as epsilon crystallin.  

Then there are some examples with duplication. Most birds have
delta crystallin. They have two closely related genes. One codes
for delta1 which is pretty lens-specific and has no enzyme
activity. The other codes for both delta2 crystallin and for the
enzyme argininosuccinate lyase (ASL). The most parsimonious
explanation is that ASL became an enzyme crystallin (like LDHB etc)
first and duplication and specialization followed. We are currently
working on a reptile which may have only ASL/delta2 and which may
therefore represent a stage before the duplication occurred.

Another example is alpha-crystallin. AlphaB is a small heat shock
protein with a probable orthologue in (eyeless) C.elegans. It is
also a major crystallin in most vertebrates. Vertebrates also have
alphaA. This has almost no non-lens expression. Again the most
parsimonious explanation is that alphaB was recruited first and
duplicated later. 
The alternative is alphaB first duplicated in an ancestor of
vertebrates producing alphaA which evidently has no important
function outside the lens. Later, both genes were recruited to the
lens independently. This is not impossible, it is just less likely.


> What is attractive about this is that the both the gain of
function
> and the subsequent duplication can have some selective component.

:If that's what you like...
:I really believe this is a point of aesthetics: in what kind of
:world do
:you like to believe biology exists.  Is it possible that the
:"classic"
:scheme is the dominant situation?  Since there's no general data
:on this
:point (your examples notwithstanding), I don't think we can answer
:this
:question as scientists.  
:What would be needed are arguments of the form, "There hasn't been
:enough
:time since the origin of life for blind enzymatic stumbling to
:have
:produced the diversity seen," and I don't believe such an argument
:can be
:forcefully motivated.  So you're left simply saying, "On aesthetic
:grounds, I don't find it plausable that there's been enough time
:for blind
:enzymatic stumbling..."  It's legitimate to feel that way, but in
:and of
:itself, it ain't science.

Well its nice to be told what is and isn't science although I think
you may be confusing aesthetics with likelihood. 
We are trying (in our humble way) to fill in as much of this large
story as we can. Much remains to be done. What I have tried to
point out is that there is at least one other perfectly reasonable
model for gene duplication which has a greater selective component
than random duplication followed by unselected sequence drift which
by chance arrives at a new function which can be properly
expressed. I am definitely not saying that this is the only way
gene duplication occurs. However, how many good examples of the
classic model are there in which it is clear that duplication and
drift occurred first?
 
 Graeme Wistow. 
 graeme at mge2.nei.nih.gov





More information about the Mol-evol mailing list